Detection of partial discharges (PDs) is widely used as a condition assessment tool for high voltage equipment. Application of low frequency test voltage is often preferred in the case of test objects with a large capacitance. The question addressed here is how results from PD-measurements performed at low frequencies correspond to that occurring at 50 Hz power frequency. Different theoretical models for void voltage were examined and compared to experiments performed on laboratory samples of mica/epoxy, including embedded cylindrical voids. All test objects were preconditioned at 10 kV and 50 Hz for 5 min before partial discharge inception voltage (PDIV) testing by stepwise increasing the test voltage from 0 to 10 kV. The PDIV test was first completed at 300 Hz before being repeated at decreasing frequencies down to 0.1 Hz. The temperature was varied in the range of 20° to 155°C. The results at high frequencies showed that a pure capacitive model fits well to the measurements. Measured dielectric response in mica/epoxy explained the decreasing PDIV at low frequencies and high temperatures. A high PDIV was measured at a combination of low temperatures and low frequencies. This was correlated with a reduced void resistance of the electrically stressed void sidewalls caused by the PD activity during the preconditioning period. This indicates that the effect of PD by-products decays faster at higher temperatures. Values of PDIV are, therefore, expected to be dependent on both temperature and frequency.
Stator winding insulation is the part of the hydrogenerator experiencing the highest number and the most damaging failures. Partial discharge (PD) measurement, both offline and online, are commonly used for condition assessment and monitoring of electrical machines. The main concern of using very low frequency (VLF) methods is the changed electrical field distributions compared to that at power frequency. Hence, PD measurements performed at VLF should be carefully assessed and compared to PD measurements at power frequencies. In this work, offline PD measurement at VLF are presented and compared to power frequency online measurements of a 50-yearold hydrogenerator in Norway, using statistical analysis of phase resolved PD recordings. It is found that both offline VLF and online assessment can identify unnormal PD activity in a specific phase, although the phase resolved PD patterns are not similar for VLF offline and online assessment.
Condition assessment of high voltage equipment based on partial discharge measurements is often performed after a voltage pre-conditioning period. The aim of this paper is to present results from experimental examinations of time variance of partial discharge activity and to propose physical explanations of the phenomena observed. Experiments were performed on laboratory made 3 mm thick discs of generator bar insulation, consisting of mica and glass fiber reinforced epoxy with a 0.5 mm thick cylindrical void surfaces of 10 mm in diameter. The effect of conducting and insulating void surfaces was examined using copper tape as upper and lower electrodes of the voids. All objects were tested by 12 one-minute long AC voltage frequency sweeps at frequencies from 50 Hz to 0.1 Hz distributed in time from start of the experiment, after initial one-hour constant 50 Hz voltage application and during object short-circuiting for 20 h. The main result shows that in case of insulating voids the apparent charges vanished after the one-hour constant 50 Hz voltage application. After a grounding period of 5 minutes, the charge magnitudes slowly increased with time until reaching steady state after about 4-8 hours. Test objects with conductive void surfaces showed such reduction in case of PD testing at 0.1 Hz only. At voltage frequencies above 10 Hz the measured PD magnitudes were found to be nearly constant, close to the expected high theoretical value. This indicate that PD by-products strongly affected the void resistivity and thereby affect the PD activity of insulating voids. These by-products are temporary and disappear with time.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.